Failure Modes and Effects Analysis (FMEA)

Transcription

1 Failure Modes and Effects Analysis (FMEA) Sasa Mutic Washington University School of Medicine St. Louis Missouri

2 Failure Modes and Effects Analysis Objectives: To motivate the use of FMEA and to provide an introduction to the application of FMEA in RT To illustrate the dependence of the results of an FMEA on the approach used and on the individuals performing the analysis FMEA is a tool 2

3 Modern RT Recent sophistication large fraction of modern treatment practices developed in the past ten years High technical complexity Multiple systems (software and hardware) non-existent guidance and regulations Limited to non High pressure Increased potential for catastrophic failures To error is human. To really foul things up requires a computer.

4 Systems Engineering The function of systems engineering is to guide the engineering of complex systems It is founded on a belief that individual components of an organization are dependent on each other It is very much about employing common sense in design of operations A set of tools for more effective management of interconnected components

5 Systems Engineering Applicable to systems with the following attributes: Complex Engineered Advanced technology High risk RT High cost Modern

6 Systems Engineering Systems Design Quality systems Human factors FMEA Systems Analysis Modeling and simulation Enterprise management Financial engineering and risk analysis Knowledge discovery Systems Control SPC Scheduling

7 Systems Engineering It is difficult for engineers to change human nature and therefore, instead of trying to persuade people not to make errors, we should accept people as we find them and try to remove opportunities for error by changing work situation. An engineers view of human error - Trevor Kletz

8 Systems Engineering in Healthcare An outline for use of Systems Engineering for improvement of national health care system We often call this arrangement a health care system even though it was never created as a system and has never performed as a system. National Academy of Engineering and Institute of Medicine, 2005

9 Failure Modes and Effects Analysis Failure Modes and Effects Analysis (FMEA) Identifies ways that a sub-process or product can fail and develops plans to prevent those failures. FMEA is especially useful with high-risk projects. The FMEA process is a dynamic,, structured approach that has the goal of linking the FAILURE MODES to an EFFECT over time for the purpose of prevention. /FMEA-&- Measurement_Systems_Analysis.ppt 9

10 Failure Modes and Effects Analysis What s s the point? Provides a structured way of prioritizing risk. Helps to focus efforts aimed at minimizing adverse outcomes. How does FMEA do that? Assembles a group of people (experts) and asks them to dream up potential risks (failure modes) and to assign a few numbers to them. Numbers are easy to sort. 10 The purpose of computing is insight, not numbers! R.W. Hamming

11 Failure Modes and Effects Analysis It can be used as a standalone tool or as a part of a broader quality system FMEA part of FDA process Applications: Equipment/product d QA/QC development Process development 11

12 Failure Modes and Effects Analysis Vocabulary Failure Mode: : How a part or process can fail to meet specifications. Cause: : A deficiency that results in a failure mode; sources of variation. Effect: : Impact on customer if the failure mode is not prevented or corrected. 12

13 Failure definition and spectrum Publicized - One side of the spectrum, usually large dosimetric errors NY Times Articles Semi-publicized RPC data Approximately 30% of participating institutions fail to deliver IMRT dose indicated in their treatment plans to within 7% or 4mm to an anthropomorphic phantom (IJROBP. 2008;71(1 Suppl):S71-5). Unpublicized/unnoted everyday occurrences Small dosimetric errors and geographic misses Suboptimal treatment plans (contouring and dose distributions) Care coordination issues Unnecessary treatment delays

14 Organizational Goals RT - Service or Manufacturing Industry? Quality Patient and employee safety Patient and employee satisfaction Efficiency 14

15 Process Itself Matters High-quality means minimizing process variation and moving the average closer to the optimum value - Med. Phys., (5): p Stable and well defined processes enable: Standardization Quantification Benchmarking Improvements Quality Control

16 FMEA Structure/Process Process Step A Process Step B Process Step C Process Step D Identify Potential Failure Mode Identify Potential Effect(s) of Failure Mode Determine Severity Identify Potential Cause(s) of Failure Mode Determine Occurance Evaluate Current Controls or Design Verification Process Determine Detectability Determine RPN Identify Actions Leading to Improvement 16

17 Failure Modes and Effects Analysis Identifying potential failure modes Must be comprehensive Must be unambiguous remove interpretation May be linked to severity Identifying potential effect(s) of failure mode Use most severe Easiest one to agree on 17

18 FMEA in Numbers Occurrence (O) describes the probability that a particular cause for the occurrence of a failure mode occurs. (1-10) 10) Severity (S) describes the severity of the effect on the final process outcome resulting from the failure mode if it is not detected or corrected. (1-10) 10) Lack of Detectability (D) describes the probability that the failure will not be detected. (1-10) 10) 18

19 RPN Score Three part system Probability of failure - O Severity of failure - S Probability that a failure would NOT be detected - D Probability of error Severity Probability no detection RPN A B C A*B*C Risk Priority Number (RPN) = Frequency * Severity * Probability 19

20 Scoring frequency of failure Qualitative Review Ranking Failure Rate Failure is unlikely 1 < 100 ppm Few failures 3 < 500 ppm Occasional failure 6 < 0.5 % Repeated failures 8 < 2.0 % Failures are inevitable 10 > 5.0 % 20

21 Six Sigma Levels of Performance Sigma Level DPMO Error as % Quality Yield Cost of Quality/Cost of Poor Quality as % of Total Operating Cost 2 308, % 69.0% Uncompetitive 3 66, % 93.3% 24-40% 4 6, % 99.4% 15-20% % 99.98% 5-15% % % World Class The Journal of Emergency Medicine, Vol. 38, No. 1, pp ,

22 Scoring Severity of Failure Severity Rank Minor to Inconvenience 1-3 Minor dosimetric error 4 Limited toxicity 5-6 Serious toxicity 7-9 Catastrophic 10 22

23 Probability that a failure will NOT be detected Probability Rank 1/10, / / / / /50 8 1/20 9 1/

24 Failure Modes and Effects Analysis NO special QA\QC QC in place for any of the processes or equipment Errors are discovered only downstream through routine processes Drastically different from all our training and beliefs Challenging to continuously remember 24

25 Failure Modes and Effects Analysis Identifying potential cause(s) of failure mode Depends on experience Often get a wide spread Evaluate current controls or design verification process Assume no QC This is difficult to do for us 25

26 Failure Modes and Effects Analysis Each failure can have multiple causes Each failure can have multiple consequences Example Isocenter misplacement Causes Laser misalignment Patient setup misinterpretation Many others.. Consequences/Severity Depend on magnitude of misplacement Use the most sever one 26

27 Failure Modes and Effects Analysis Two major challenges in completing an FMEA 1.To be confident that all possible significant failure modes have been identified. This requires that experts contributing to the FMEA have a wide range of experience 2.The description of the failure mode must be completely clear and different sources that result in the same failure must be differentiated. Different sources may result in the same failure but will have different likelyhoods of occurrence. 27 ICRP Publication Preventing Accidental Exposures from New External Beam Radiation Therapy Technologies

28 AAPM - TG100 FMEA Application of Risk Analysis Methods to Radiation Therapy Quality Management 28

29 TG100 IMRT Example STEP 1. Process Steps 1. Diagnosis, Staging, History and Physical 2. Patient entered in database, assigned db keys, etc 3. Decision to Treat with Radiation 4. Scheduling for Planning Process 5. Immobilization and Positioning 6. Each Imaging Procedure (CT, MR, PET ) 7. Transfer Images 29

30 TG100 IMRT Example STEP 1. Process Steps 8. Initial Treatment Planning Directive (from MD) 9. RTP Anatomy 10. MD Review 11. Plan Review 12. Plan Approval 13. Plan Prep 14. IMRT QA 15. Clinical Plan Prep 16. Day 1 Treatment: 17. Day n Tx: 18. Weekly Chart Check: 19. Therapy Complete 30

31 TG100 IMRT Example Single Step - Subcategories 9. RTP Anatomy : Import images into RTP system db Create case (anatomy) Import appropriate image datasets into case Dataset registration (fusion) of multiple datasets Delineate targets (MD) and normal structures for planning, optimization Complete anatomical description (objects, fusion, 3-D D review, etc) 31

32 TG100 IMRT Example STEP 2. Process Map Immobilization and positioning Transfer patient data to treatment delivery Clinical plan prep Manual data entry and plan modification Immobilization equipment fabricated MD review Specify treatment course Immobilization equipment documented, labeled, and stored Check anatomy Plan review by dosimetrist Delivery protocols Scheduling Chart filing Patient database information entered Data into electronic database RTP anatomy contouring Prescription Check that previous treatments were accounted for Check normal tissue and target goals, dose distributions Physics Check Equipment Check Data into written chart Import and fuse images MD: delineate GTV Delineate ROIs and planning structures Check dose calc method Check beams, machines, etc. DQA Chart Check Imaging and diagnosis Edit density map for artifacts PTV construction Plan approval Review of patient medical history Weekly chart checks Successful treatment Decision of protocol Decision of treatment technique Scheduling for planning process Immobilization equipment documented, labeled, and stored Immobilization for Imaging study Immobilization equipment fabricated Patient prepped (contrast, tattoos, BBs etc.) Set up data documented Specify images for target/structure delineation Specify protocol for delineating target and structures Specify registration goals Indicate motion/uncertainty management Suggest initial guidelines for treatment parameters Specify dose limits and goals Construct Blocks Compensators Bolus, Etc Dosimetrist check Plan preparation Documentation Treatment Dosimetry Physics Positioning Scheduling MD Subsequent treatments Treatment review Consultation and decision to treat Enter prescription Setup fields Setup dose calc parameters Optimization Time out Imaging (CT/PET/MR) Evaluate plan Positioning Treatment planning Documentation 32 Transfer images and other DICOM data: primary and secondary data transferred Initial treatment planning directive Initial treatment Imaging (port films, CBCT, etc) Treatment

33 TG100 IMRT Example Failure Modes Process Number of Failure Modes Patient Entered into database 3 Immobilization and positioning 7 CT Simulation 13 Other imaging studies 7 Transfer of images and other DICOM data 8 Initial treatment directive from MD 9 Define anatomy for treatment planning 31 33

34 TG100 IMRT Example Failure Modes Process Number of Failure Modes Treatment planning 53 Plan approval 11 Prepare plan for delivery 30 First session treatment 20 Subsequent treatments 23 34

35 35 Table Example

36 TG100 IMRT Example STEP 3. FMEA Three part system Probability of failure - O Severity of failure - S Probability that a failure would NOT be detected - D Probability of error Severity Probability no detection RPN A B C A*B*C Risk Priority Number (RPN) = Frequency * Severity * Probability 36

37 TG-100 OSD Scores Nine of the TG members assigned O, S, D values Average OSD and RPN values and corresponding standard deviations created Highest and lowest median RPN scores used to identify most and least critical steps 10% and 20% values in both directions used as threshold Result is High risk steps (HM10 and HM20) and Low risk steps (LM10 and LM20) Steps with severity higher than 8 are given special priority regardless of RPN value 37

38 Your OSD Scores 38 Start with TG-100 scores VERY GOOD LEARNING TOOL Start with blank process and develop your own program Involve multiple groups Administrative Dosimetry Physics Clinical (Physicians) Therapy Nursing Develop individual OSD and RPN scores and analyze the data

39 Interactive Example Incorrect Capping Step Potential failure mode Evaluate ROIs (targets and structures) Incorrect ROI volumes Cause Inadequate design/programming Potential Cause Inadequate contour capping algorithm (3D expansion) Potential Effects Wrong dose, suboptimal plan 39

40 Interactive Example Incorrect Capping Frequency O 1 to 3 3 to 7 7 and higher Severity S 1 to 3 3 to 7 7 and higher Detectability D 1 to 3 3 to 7 7 and higher 40

41 Interactive Example Incorrect Capping TG-100 Scores Probability of error Severity Probability no detection RPN Mean SD

42 TG-100 Lessons Scoring overall inconsistent with unclear trends High and low RPN values scored much more uniformly Example High RPN: Failure to notice that MLC leaves are not moving O (3.33 ± 0.87), S (8.22±1.2), D (8.22±1.56) and RPN (221.7 ± 64.8) 42

43 STEP 5/6. Process Map Marking Shows if potential failures are uniformly distributed through the process or clustered If clustered, should consider the major step as a hazard Design QM processes according to clustered areas and severity scores 43

44 44 Process Map Marking

45 Conclusions FMEA can be used as a standalone effort or as a part of a broader process improvement effort Best performed with all involved groups (content experts) Need a facilitator lead person with expertise in FMEA A broad spectrum of outcomes, from better insight and understanding to a much better product or service 45